phosphorus removal by blast furnace slag and cement...
TRANSCRIPT
300 © IWA Publishing 2011 Water Quality Research Journal of Canada | 46.4 | 2011
Phosphorus removal by blast furnace slag and cement
clinker – flow cell studies for estimation of sorptive
capacity for use with constructed treatment wetlands
Anamika Sikdar Paul and Bruce Anderson
ABSTRACT
Blast furnace slag and cement clinker were explored in long-term flow cell experiments for
estimation of their phosphorus (P) removal efficiencies. A local gravel, typically used in constructed
treatment wetlands, was used as a control medium. The experiments examined the removal of
phosphorus from a solution initially containing 4 mg P/L. The slag and clinker were nearly 100%
efficient due to very high sorptive capacities. The control gravel medium removed 50% of the influent
phosphorus. Results from this study indicate that the use of blast furnace slag in constructed
wetlands or filter beds is a promising solution for P removal via sorption mechanisms.
doi: 10.2166/wqrjc.2011.112
Anamika Sikdar Paul (corresponding author)AMEC Earth and Environmental,160, Traders Blvd (E),Suite 110, Mississauga, ON,L4Z 3K7 CanadaE-mail: [email protected]
Bruce AndersonDepartment of Civil Engineering,Queen’s University,Kingston ON,Canada,K7L 3N6 Canada
Key words | adsorption, clinker, constructed wetland, phosphorus, slag, sorption capacity
INTRODUCTION
The adverse effects of eutrophication due to the presence of
phosphorus in surface waters are well established (Orive
et al. ). The OntarioWater Resources Act 1990 Guideline
F-5 sets the total phosphorus limit of 1 mg/L formunicipal and
private sewage treatment systems discharging into a water-
body. Sunny Creek Estates (SCE) is a mobile home village
using a combination lagoon–CW system to treat its sewage
and discharges directly into the Bay of Quinte. The Bay of
Quinte on the north eastern shore of Lake Ontario is a recog-
nized Area of Concern. The Bay of Quinte Remedial Action
Plan (BQRAP) set an objective of 0.3 mg/L of total phos-
phorus and a further stringent guideline of 0.1 mg/L is
presently being considered. Currently, the SCE treatment
system does not achieve compliance with respect to effluent
concentrations of phosphorus, and better treatment is needed.
Conventional technologies for removal of phosphorus
from point source wastewater discharges are physical pro-
cesses (settling, filtration), chemical precipitation (with
aluminum, iron and calcium salts) and biological processes
that rely on biomass growth (bacteria, algae, plants) or
intracellular bacterial polyphosphates accumulation
(Bashan & De-Bashan ). Long-term studies and
increased operational experience indicate that phosphorus
removal is variable or inconsistent (Richardson & Craft
; Reed & Brown ; Wood ) in subsurface con-
structed wetlands (CW) that can be attributed to the
complexity of phosphorus removal mechanisms, and the
lack of consideration of these complexities in design.
The major factors that make P removal by the wetlands
particularly difficult are the type, quantity and diversity of
the influents that need to be treated. The principal phos-
phorus removal mechanism, adsorption/precipitation,
being a finite process, requires the P saturated substrate to
be replaced after a certain operational period (Faulkner &
Richardson ; Mann & Bavor ; Drizo et al. ;
Shilton et al. ). Given these, the sorption and deso-
rption of phosphorus in constructed wetlands is impacted
not only by the physical and/or chemical characteristics of
the substrate media, but also by phosphorus loading, hydrau-
lic conditions, temperature, time and dissolved oxygen.
When designing a CW for P removal, the selection of the
material to be used as thewetland substrate (rootingmedium)
plays a crucial role (Mann & Bavor ; Drizo et al. ,
; Johansson & Gustafsson ). A potential medium
301 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
for a constructed wetland, if rich in Fe, Al and Ca ions, would
be considered for application in the treatment process. A gen-
eral summary of the media mentioned in the literature as
having been tested as wetland soils are blast furnace slag,
light expandable clay aggregates, cement clinker, crushed
brick, activated aluminum oxide, half burned dolomite,
maerl, siliceous sedimentary rock opaka, peat amended
with steel wool, red mud, mussel shells and shale.
When tested in batch and column laboratory experiments
at both low and high P concentrations, blast furnace (BF)
slags, waste products from iron production demonstrated
high P removal efficiencies (Mann & Bavor ; Baker
et al. ; Johansson & Gustafsson ). The results
observed by other researchers (Drizo et al. ) confirmed
that steel slag showedexcellent removal efficiencywhen com-
pared to the other investigated materials. Rosolen ()
tested 17 different materials in batch tests, and it was noted
that slag material and clinker showed the highest potential
of achieving efficient P removal. Van Weelden () and
Laska () evaluated P removal efficiencies by zebra
mussel shells, which are primarily made up of calcium com-
pounds, demonstrating satisfying sorption potential.
Although batch tests provide a good indication of the
material’s capacity to retain P, when making a selection of
potential substrates to be used in designing constructed wet-
lands and/or filters for P removal it is necessary to conduct
long-term column or pilot-scale experiments (Drizo et al.
; Shilton et al. ). These will allow evaluation of
the influential factors for both sorption and desorption in
these test systems.
The main focus of this research was to develop and
implement a program of experimental investigations into
the behavior of phosphorus sorption in the media surface
under normal laboratory conditions representative of those
occurring in the field. The work would sufficiently charac-
terize the media for eventual application in a filter or as
the rooting media in the wetland.
The research program was conducted in three stages.
The first phase involved conducting batch studies and inves-
tigating the reaction kinetics of the batch studies. The last
phase of the program was conducted to investigate the
forms of sorbed phosphorus within the media. The results
of the first and last phase of the program are being presented
in separate papers. The present paper presents the results of
the second phase of the research program. During the
second stage of the research program, long-term column
experiments using custom designed flow cells were con-
ducted for estimating the P removal efficiencies of slag
and clinker media. In addition, local gravel was used as a
control medium and the amount of P retained by each
medium was determined.
MATERIALS
Sample preparation
The media were washed, sorted and dried prior to commen-
cing formal laboratory investigations. Emphasis on local
availability, amount of processing required, natural abun-
dance and cost effectiveness were considered during
media selection. The slag was obtained from National Slag
Limited, Hamilton and clinker from Essroc Cement,
Picton. Gravel used as a control medium was being used
in the wetland in SCE.
The size used for all tests was 0.5–8 mm. The size ranges
for the media reported in the literature are 2.5–10 mm for
slag (Drizo et al. ), 2–10 mm for fine gravel (Seo et al.
), 8–16 mm for gravel (Prochaska & Zoubilis 2006),
2–5 mm for very fine gravel (Xu et al. ), 0.2–3.2 mm
for sand (Arias et al. ) and 3.5–10 mm for gravel
(Garcia et al. ). As such, the size range used here allowed
for some comparison to the previous studies. An unprocessed
(unwashed and not sorted) fraction of each material was
retained for subsequent media characterization studies.
METHODS
Laboratory-scale experiments were conducted to examine
the physical and hydraulic properties, and the phosphorus
retention mechanisms, of the three candidate media.
Media characterization
Thehydraulic and geometric designof thewetlandfilterwill be
strongly influenced by the physical properties of the selected
medium. Information on media particle size distribution,
302 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
along with porosity and hydraulic conductivity, can be used to
predict the movement of water through the filter.
Standard laboratory tests were performed to determine
the following:
• Particle size distribution analysis (ASTM D 422 )
• Bulk density determination (Black )
• Particle density determination (ASTM D 854 )
• Porosity calculations (Vomicil )
• Soil alkalinity (pH) measurements (Clesceri et al. ;
ASTM D 4972 )
• Hydraulic conductivity determination (ASTMD5856 )
Hydraulic conductivity
Constant head tests were performed using an up-flow per-
meameter to determine the hydraulic conductivity (K)
values of the media, as shown in Figure 1. The media were
placed in the permeameter and lightly compacted. For
each test, K was determined using Darcy’s Law for uncon-
fined flow through a porous medium as outlined in
Equation (1) (Reed & Brown ):
K ¼ VLAht
(1)
where V is the fluid volume (m3), K is the hydraulic conduc-
tivity (m/s), h is the differential head across the sample (m),
Figure 1 | Constant head test apparatus and permeameter for hydraulic conductivity
tests.
L is the sample length (m), A is the cross-sectional area of
media sample (m2) and t is the time (s).
Horizontal flow cell experiments
Batch testing is a convenient procedure for comparing the
performance potential of various sorbent materials; how-
ever, bench scale test results are only indicative of the
performance under static test conditions, and should not
be extrapolated to model field performance (Benefield
et al. ). To overcome some of the limitations in data
interpretation and application inherent to batch shaker test-
ing, flow cell experiments were conducted using horizontal
flow filter cells to investigate phosphorus retention proper-
ties of gravel, slag and clinker media.
Procedure
Figure 2 shows the testing apparatus used in the flow cell
studies. Approximately 5,100 cm3 of the media was placed
in the middle chamber of a three chamber plexiglass hori-
zontal flow cell. Figure 3 shows the dimensions of a flow
cell and the location of sampling wells. The total length of
each flow cell was 775 mm and the length of the middle
chamber was 600 mm. The lengths of the inlet and outlet
chambers were 110 and 65 mm, respectively. The height
and the width of the flow cell were 95 and 100 mm,
respectively.
An influent flow of 10 L/day was maintained continu-
ously throughout the study. The influent P concentration
Figure 2 | Horizontal flow filter cells (from left: slag, gravel and clinker).
Figure 3 | Details of the flow cell.
303 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
of 4 mg/L (±0.001) was passed continuously through the
flow cells. The influent concentration was based on
observed wetland influent concentrations and those noted
in the literature, which ranged from 30 to 1 mg/L (Drizo
et al. , , ; Johansson & Gustafsson ;
Arias et al. ; Metcalf & Eddy Inc. ).
At regular intervals, liquid from the sampling wells in
the flow cell was collected from fixed sampling locations
during the test. For the gravel flow cell, samples were col-
lected from location A (at 10 cm), B (at 25 cm), C (at
40 cm) and D (at 50 cm) from the inlet. For the slag and clin-
ker flow cells, the samples were collected from location A
(at 15 cm), B (at 30 cm) and C (at 45 cm) from the inlet.
The samples were analyzed for orthophosphate by
Quickchem method 10-115-01-1-A (Diamond ).
Samples were collected at intervals of 3 days for gravel
and 4 days for slag and clinker. For each sampling period,
incremental treatment volumes were calculated by multiply-
ing the flow rate by the length of the interval period. After a
period of 10 weeks, the flow cell study was terminated.
Treatment profiles
The phosphorus concentration profiles in the liquid phase
were developed along the length of each filter cell, and rep-
resented the amount of phosphorus being sorbed within the
media as time passed.
Sorption capacity
Orthophosphate removal was calculated based on percent
removal, R (%), of the incoming concentration. The phos-
phorus removed during a sampling interval period was
calculated based on the total load delivered to the filter
since the last sampling period and is given as follows:
P ¼ R100
� (Vi � Vs(i�1)) �Cin (2)
where P is the total phosphorus load (mg), Vi is the volume
treated to date (L) and Vs(i�1) is the total volume treated at
the previous sampling period (L) and Cin is the influent P
concentration (mg/L) in flow cell.
A section of the flow cell, Ms (kg) is the region between
the two adjacent sampling wells. The mass of a section of
media was estimated by using the percentage of the total
length of the filter in relation to the total mass of the filter.
The removal of phosphorus per unit mass for a given
time period, Xi (mg/kg), was the ratio of P to Ms. Finally,
the cumulative phosphorus sorptive capacity of each filter
cell was calculated in each section of the cell.
RESULTS AND DISCUSSION
Media characterization
Particle density (ρs), bulk density (ρb), porosity (η), hydraulic
conductivity (K) and soil (pH) for the three media are sum-
marized in this section.
Grain size distribution
Figure 4 shows the grain size distribution of the three media
plotted as %-finer on a grain size distribution curve. The
majority (∼70%) of the gravel was retained on the smaller
sieves (<2 mm) whereas the majority of the slag (∼82%)
Figure 4 | Plot of %-finer of media grain particles as a function of sieve size.
Table 1 | Physical characteristics of candidate media
Bulkdensity
Particledensity Porosity Hydraulic
304 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
and clinker (∼77%) were retained on the larger size
(>2 mm) sieves. Gravel had an effective grain size (d10) of
0.5 mm, coefficient of uniformity (Cu) of 3.40 and a coeffi-
cient of curvature (Cc) of 0.75. Clinker had an effective
grain size of 0.9 mm, coefficient of uniformity (Cu) of 4.44
and a coefficient of curvature (Cc) of 1.74. Slag had an effec-
tive grain size of 1.2 mm, coefficient of uniformity (Cu) of
2.92 and a coefficient of curvature (Cc) of 1.37. Coefficient
of curvature (Cc) and coefficient of uniformity (Cu) are
derived from effective grain sizes d10, d30 and d60.
All three media are classified as poorly sorted sand (SP)
under the Unified Soil Classification system based on the
values of Cc and Cu (ASTM D 2487 ). Recent Danish
guidelines as mentioned in Arias et al. () recommend
a d10: 0.3–2 mm, d60: 0.5–8 mm and Cu< 4 to ensure ade-
quate hydraulic conductivity. Except clinker (Cu¼ 4.44) all
other criteria were satisfied by the media. All three media
were expected to show good performance for phosphorus
sorption, based on studies such as that of Xu et al. (),
who noted that the phosphorus sorption capacity was influ-
enced by the physico-chemical characteristics of the media,
wherein finer grain size resulted in higher phosphorus
sorption.
Medium (g/cm3) (g/cm3) (%) pH conductivitya (m/s)Gravel 1.603 2.6737 40.03 7.8 0.0056 (±0.001)
Slag 1.163 2.5874 55.06 10.8 0.25 (±0.07)
Clinker 1.004 3.1460 68.08 11.4 0.19 (±0.02)
aAverage of four tests.
Density and porosity
The observed results from the bulk density, particle density
and porosity values were consistent with values reported
by Rosolen (). Table 1 summarizes the average values
for the bulk density, particle density, porosity, pH and
hydraulic conductivity for the three media. Porosity affects
the fluid flowrate through the media, and a higher porosity
may correspond to a greater surface area and, therefore,
more exposed sites for phosphorus adsorption. The design
range for porosity recommended by Kadlec & Knight
() is 35–45%. Clinker was the most porous material.
Gravel was observed to have the recommended design por-
osity whereas clinker and slag had greater porosity than
recommended for use as a wetland medium for phosphorus.
However, as is discussed later, the high porosity values of
clinker and slag did not act in a detrimental way for phos-
phorus removal.
Hydraulic conductivity
Hydraulic conductivity tests showed that slag had the
highest conductivity, followed closely by clinker, and
Table 2 | Hydraulic loading parameters and effluent pH for flow cells
MediumPorosity(η)
Linear fluidvelocity, vx,(cm/h)
Hydraulicresidence time(h) Rea pH
Gravel 0.40 12.30 4.88 0.05 7.4–7.6
Slag 0.55 8.94 6.71 0.08 9.3–10.5
Clinker 0.68 7.23 8.30 0.06 9.5–11.0
aCalculated using linear fluid velocity.
305 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
finally the gravel. The recommended design range for the
hydraulic conductivity for the constructed wetlands
medium given by Reed & Brown () is 0.0012–0.12 m/s
(100–10,000 m3/m2 per day). Gravel had a hydraulic con-
ductivity of 0.0056 m/s which is within the recommended
design range given by Reed & Brown (). Even though
the hydraulic conductivities of clinker and slag were
higher than the recommended range, they still proved to
be good media for phosphorus removal as noted later.
This was probably due to the fast reaction times of the phos-
phorus sorptive reactions, which would be beneficial in
downsizing the filter bed.
Though the laboratory studies use consistent and repro-
ducible testing techniques, field conditions seldom resemble
laboratory conditions. Hydraulic conductivities calculated
in the laboratory could be reduced by as much as a factor
of ten in the field (Kadlec & Knight ). Calder et al.
() reported that the hydraulic conductivities at their
field site were reduced considerably, by a factor greater
than 10, from the laboratory derived design values.
pH
pH is an important factor for the phosphorus reactions as it
controls the type of chemicals taking part in the reactions
and also affects the dissolved oxygen content of the receiv-
ing water body indirectly (Drizo et al. ; Reddy &
D’Angelo ; Garcia et al. ). The effluent from the
gravel flow cell was able to meet the regulatory pH require-
ment whereas the effluent from the slag and the clinker flow
cell had high pH and did not meet the regulatory pH
requirement of 6–9.5 (Environmental Protection Act –
O. Reg. 560/94, Ontario Ministry of Environment ).
Thus, it may be necessary to have a neutralization process
(e.g. secondary filter with a low pH media like acidic peat)
after the application of slag or clinker for sorption in order
to regulate the final effluent pH.
The physical characteristics all influence the effective-
ness of a medium for the removal of phosphorus (Drizo
et al. ; Prochaska & Zouboulis ). However, these
effects are usually on a short-term basis. For long-term phos-
phorus removal to be enhanced and sustainable, additional
properties (e.g. chemical composition) of the medium and/
or operational parameters of the wetland or filter become
important (Arias et al. ). The additional properties and
operational parameters can be evaluated using flow cells
as described in the next section.
Horizontal flow cell experiments
Horizontal flow cell experiments were conducted to investi-
gate the phosphorus retention properties of the three media
(gravel, clinker and slag).
Hydraulic loading parameters
The linear fluid velocity (vx) and hydraulic residence time
(HRT), which are dependent on the porosity, were calcu-
lated. Table 2 summarizes the hydraulic residence time,
linear fluid velocity, Reynolds number (Re) in the flow
cells, the porosity of the media and the range of pH at the
outlet of the flow cells.
The linear fluid velocity was different in each flow cell,
due to the different porosity of each medium. The flow
cell tests were conducted using a constant flow rate of
10 L/day, which resulted in the hydraulic residence times
of a few hours. The focus of this research was to have com-
parable loading conditions (10 L/day), to allow comparison
of the tests results between the media. Any field setup using
residence times similar to that used in the present work (e.g.
10 h) should remove comparable amounts of phosphorus
from the wastewater.
The retention times of slag and clinker are greater than
those of gravel, but even if the residence time was increased
for gravel, the performance probably would not change due
to the inherently poorer sorption characteristics of the
medium. Rosolen () used a hydraulic retention time of
approximately 1 day. Calder et al. () had approximately
306 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
6 days retention time for similar gravel; even then the gravel
(comparable medium to the present study) performed poorly
(P removal – 25%) as compared to the slag (P removal –
65%). Interestingly, Toet et al. () reported that the
increase of hydraulic retention times from 0.3 to 9.3 days
did not increase the phosphorus removal. Based on the
above studies, for the present study, hydraulic retention
time was considered not to be an influential factor for sorp-
tion. Future work focusing on the study of hydraulic
retention times is required to confirm the assumption. The
low values of Reynolds number (Re< 1) shows that the
flow is Darcian in nature. The pH observed in the flow
cells shows that the outflow from the slag and clinker are
alkaline in nature.
Phosphorus sorption profiles
Soluble phosphorus concentrations were monitored at the
sampling locations by collecting and analyzing samples
from these locations in each cell. As mentioned earlier, for
the gravel flow cell, samples were collected from location
A (at 10 cm), B (at 25 cm), C (at 40 cm) and D (at 50 cm)
Figure 5 | Phosphorus sorbed by candidate media flow cells.
from the inlet. For the slag and clinker flow cells, the
samples were collected from location A (at 15 cm), B (at
30 cm) and C (at 45 cm) from the inlet.
Phosphorus sorption profiles (mg of P sorbed per L of
influent solution) were developed for each flow cell, and
are shown in Figure 5. The outflow phosphorus concen-
tration for the gravel flow cell stabilized at approximately
1.75 mg/L after 2 weeks which indicates that the concen-
tration of sorbed phosphorus was 2.25 mg/L. The same
trend was noted for the samples taken from all four
sampling locations in the cell. The amount of phosphorus
sorbed was higher close to the inlet than in the later sections
of the flow cell. This was not unexpected as phosphorus was
steadily being sorbed along the length of the flow cell, leav-
ing less phosphorus in the solution in the latter sections. The
gravel showed good initial performance until about 20 days,
after which all concentrations were similar. Although the
gravel removed approximately 50% of the influent phos-
phorus, it did not meet the regulatory requirement of
1 mg/L. It should be noted that gravel was used as a control
medium for studying the hydraulic properties and was not
really expected to be a good medium for phosphorus
307 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
removal. A similar sourced gravel used by Rosolen () in
her work also showed good performance initially (until
about 15 days) and this medium achieved equilibrium as
indicated by a constant concentration of phosphorus in
the effluent. Near saturation (effluent values 90% of influ-
ent) was noted by 36 days.
In the case of slag, the sorbed phosphorus was more
than 3.95 mg/L during the entire test period as indicated
by the outflow concentration of less than 0.05 mg/L.
During the initial period, the concentration of sorbed phos-
phorus fluctuated slightly in the first section but, after 20
days, it was more than 3.5 mg/L consistently until the con-
clusion of the test. The sudden performance improvement
in the first section (at around 20 days), unlike anything men-
tioned in the literature, was unanticipated. This
performance improvement would be significant, with the
filter now being compliant in the first section as well.
Although this initial high effluent concentration of phos-
phorus (approximately 1.5 mg/L) is not a matter of
concern as this happened only in the first section, it does
warrant some discussion. This could be caused by fines
being washed out of the filter, and consequent opening up
of the micropore sorption sites. This might become impor-
tant if one would like to shorten the length of the filter for
other applications (e.g. to optimize capital costs). This may
indicate that an acclimation period might be required for
shorter slag filters in the field before they become compliant.
However, the need for the acclimatization period can not be
really explained or confirmed by the information in the lit-
erature. It is important to note that the flow cell outflow
concentration was not effected during the test period
(<0.05 mg/L) and remained below the recommended limit
of 0.3 mg/L (the Bay of Quinte objective) and 0.1 mg/L
(the proposed Bay of Quinte objective).
For the clinker flow cell, the concentration of outflow
phosphorus was less than 0.02 mg/L during the entire test
period and in all sections, showing that the amount of
phosphorus sorbed was more than 3.98 mg/L. There was
a small discrepancy observed during the test at day 40,
but given the high amounts of sorbed phosphorus
observed, this was not considered of any consequence.
The outflow phosphorus concentration was even below
0.1 mg/L (the proposed Bay of Quinte objective) for clin-
ker. The system never achieved saturation during the test
period. The effluent phosphorus concentration from the
first section was very low, which indicates that most of
the phosphorus was being sorbed in this section.
Clinker was considered the best medium of the three
tested, as this flow cell had excellent overall performance
shown by the consistently low phosphorus outflow
concentrations.
Saturation is said to occur when the effluent concen-
tration reaches 95% of the influent concentration
(Benefield et al. ). In the present study, saturation was
never reached during the entire test period for all media,
due to the high sorption capacity of the media. As men-
tioned earlier, Rosolen () found that by increasing the
influent concentration from 8 to 22 mg/L, her gravel finally
achieved saturation (outflow concentrations of ∼21 mg/L)
within 1 week. The slag and clinker used in the same
study did not reach saturation even at the high dosing.
Drizo et al. () reported that steel slag was nearly
100% efficient in removal of phosphorus from an influent
concentration of 20 mg/L for 114 days. After increasing
the influent concentration to 400 mg/L for 21 days, the
steel slag still did not reach saturation. Higher phosphorus
concentrations such as 400 mg/L are not representative of
the typical influent concentrations in the field for domestic
wastewater applications, and this level of dosing would
not be useful for estimating filter performance. Therefore,
increasing the influent phosphorus concentration to achieve
saturation of the media was not conducted in this research.
None of the filters in the present work reached satur-
ation that leads to an inability to predict the ultimate
capacity (x/m)u of the media. Calder et al. () also
observed that the field filters never reached saturation. It
would therefore be difficult to predict the longevity of the fil-
ters and only an approximation could be made as discussed
in the following section.
Sorption capacity
The cumulative amounts of phosphorus removed per unit
mass (x/m, mg/kg) by the three media are plotted over
time in Figure 6. All three media show higher amounts of
phosphorus sorbed in the first section. This leaves less phos-
phorus in the latter sections for subsequent removal. In the
gravel and clinker flow cells, there was no difference in
Figure 6 | Phosphorus retained by sections of the flow cell media.
308 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
the mass sorbed between the latter sections, whereas in the
slag flow cell there was a small difference in the latter
sections.
Figure 6 shows that clinker was the best of the three
media, as it retained the most phosphorus (maximum
1,500 mg/kg) by the end of the test period, with slag follow-
ing close behind (1,200 mg/kg). Gravel sorbed a maximum
of approximately 700 mg/kg of the influent phosphorus
(less than 50% compared to clinker) and, as mentioned,
this medium failed to meet the regulatory requirements for
effluent phosphorus compliance concentration. In other
words, the phosphorus retained at the end of the test could
be considered as a very conservative estimate of the sorption
capacity of the medium, since none of the flow cells reached
saturation levels. In the first section, the removal efficiency of
clinker was >99% whereas for gravel it was approximately
50–60%. Slag showed a removal efficiency of >98%. These
results compare favorably with other studies in the literature.
Prochaska & Zouboulis () noted that gravel had
low phosphorus sorption capacity whereas their slag had
an extremely high sorption capacity. Drizo et al. ()
noted that their slag removed >99% of the influent
phosphorus. There have not been many studies on cement
clinker as a medium for phosphorus removal, although
some studies using coal ash demonstrated high phosphorus
removal efficiency (Kirk et al. ; Gray & Schwab ).
Arias et al. () noted in a study of 13 Danish sands (com-
parable to the gravel in the current study) that the removal
efficiency of sands all decreased significantly to levels
<50% after 12 weeks.
Rosolen () calculated the ultimate sorption
capacities, (x/m)u (mg/kg), based on the batch isotherm
tests for all three media. The (x/m)u for gravel was 15 mg/kg,
for slag it was 9,361 mg/kg and for clinker it was
14,438 mg/kg, respectively, for an influent concentration
of 8 mg/L. When compared to the conservative sorptive
capacity of gravel estimated in the present study, it was
higher than those estimated by Rosolen (). Therefore,
the sorption capacities for slag and clinker would be
expected to be very high compared to gravel and to the esti-
mated isotherm sorption capacities by Rosolen.
Since the flow cells did not achieve saturation in the pre-
sent study, a very conservative estimate of the volume of
medium required to achieve treatment for one year for the
309 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
slag and clinker is shown in Table 3. The amount of media
needed to reach the desired concentration of 1 mg/L for a
slag medium filter was assumed to be the mass present in
the first two sections, whereas for clinker it was assumed
to be the mass present in the first section only. The
measured concentration of the effluent was below 1 mg/L
for slag after the second section (<0.23 mg/L) and for clin-
ker after the first section (<0.02 mg/L). The observed
effluent concentration from the gravel flow cell was
∼2 mg/L gravel at the conclusion of the test and this was
used for the calculation of the volume of the medium.
Rosolen () also calculated a conservative estimate
of the filter volumes for the three media for an influent con-
centration of 8 mg/L. As her slag and clinker flow cells did
not reach saturation, the sorption capacity was recorded as
the highest value of sorption capacity achieved after the first
section of the filter for the slag and the clinker. The esti-
mated volume for gravel was 8,260 m3, for slag it was
54 m3 and for clinker it was 8 m3, respectively.
Table 3 indicates that the slag and clinker were much
better media with high sorption capacities, as compared to
gravel, for phosphorus removal. Despite the conservative
nature of these estimates, both slag and clinker demonstrated
excellent potential for use in the design of a reasonably sized
filter. The costs of filter replacement and spent filter disposal
or regeneration should be considered in any design exercise,
and the smallest filter with the longest lifespan would be the
most effective choice. Water quality considerations (e.g. pH
of the effluent) would also need to be addressed in the overall
impact assessment of the filters.
The criteria to evaluate the system should include the
estimated sorption capacity of the medium along with the
required effluent concentrations, influent flow rates and con-
centration (and expected variability), physical hydraulic
characteristics or a combination of these operational factors.
Table 3 | Sorption capacity and estimated filter volumes
Medium Pinitial Peffluent
Sorption capacityfrom flow cells(mg/kg)
Volume of mediafor one yearoperation (m3)
Gravel 4 mg/L ∼2 mg/L 700 193
Slag 4 mg/L <1 mg/L 1,200 130
Clinker 4 mg/L <1 mg/L 1,500 65
CONCLUSIONS
In conclusion, blast furnace slag and cement clinker showed
very high efficiencies (nearly 100%) in removal from an influ-
ent (4 mg P/L) over a 9-week period. The slag medium
accumulated 1,200 mg P/kg of slag whereas cement clinker
accumulated 1,500 mg P/kg of clinker, presumably through
the processes of specific adsorption onto metal hydroxides
and precipitation as hydroxyapatite. The maximum amount
ofP that could be removed byeithermediawasnot determined
but would exceed the values mentioned earlier. The control
medium, local limestone gravel accumulated 700 mg P/kg of
gravel. The maximum amount of sorption in this gravel
would be close to the value of 700 mg P/kg of gravel since
this medium appeared to reach a quasi-equilibrium condition.
Theoutflowconcentrationswere less than0.3 and0.02 mg P/L
from the slag and clinker flow cells, respectively. However,
the pH of the outflowwas alkaline in both flow cells. The out-
flow concentration from the gravel flow cell was 1.75 mg P/L
and the pH of the outflow was neutral. Overall, gravel
removed far less as compared to either clinker or slag media
and did not meet the discharge guideline of 1 mg P/L. Slag
and clinker showed very high sorption capacities and high
outflow pH values. Mixing peat with either media could
reduce the pH of the outflow without affecting the sorption
capacities. Also, it is expected that slag and clinker will have
an impact on the effluent quality with respect to heavy metals
present in the wastewater. Therefore, it is recommended that
future studies should investigate the impact of slag and clinker
on the removal of heavy metals from wastewater.
For design purposes, flow cell studies provide better and
more realistic estimates of ultimate sorption capacities than
do isotherm tests. Simple jar testing (e.g. isotherm testing) is
not enough to select the best medium. Flow cell testing simu-
late (in the small scale)field filter operation and performance.
These tests are important to predict the sorption capacity of
the medium for the long-term removal of phosphorus.
ACKNOWLEDGEMENT
The financial assistance provided by the Natural Sciences
and Engineering Research Council of Canada (NSERC) is
gratefully acknowledged.
310 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
REFERENCES
Arias, C., del Bubba, M. & Brix, H. Phosphorus removal bysands for use as media in subsurface flow constructed reedbeds. Water Res. 35 (5), 1159–1168.
ASTM D 2487 Standard Classification of Soils forEngineering Purposes (Unified Soil Classification System).Annual Book of ASTM Standards. 04.08. ASTM, WestConshohocken, PA, USA, pp. 206–215.
ASTM D 422 Standard Classification of Soils for EngineeringPurposes (Unified Soil Classification System). Annual Bookof ASTM Standards. 04.08. ASTM, West Conshohocken, PA,USA, pp. 10–16.
ASTM D 4972 Standard Classification of Soils forEngineering Purposes (Unified Soil Classification System).Annual Book of ASTM Standards. ASTM, WestConshohocken, PA, USA.
ASTM D 5856 Standard Classification of Soils forEngineering Purposes (Unified Soil Classification System).Annual Book of ASTM Standards. ASTM, WestConshohocken, PA, USA.
ASTM D 854 Standard Classification of Soils for EngineeringPurposes (Unified Soil Classification System). Annual Bookof ASTM Standards. ASTM, West Conshohocken, PA, USA.
Baker, M. J., Blowes, D. W. & Ptacek, C. J. Laboratorydevelopment of permeable reactive mixtures for the removalof phosphorus form onsite wastewater disposal systems.Environ. Sci. Tech. 32, 2308–2316.
Bashan, Y. & De-Bashan, L. E. Recent advances in removingphosphorus from wastewater and its use as fertilizer. WaterRes. 38 (19), 4222–4246.
Benefield, L. D., Judkins, J. F. & Weand, B. L. ProcessChemistry for Water and Wastewater Treatment. Prentice-Hall Inc, Englewood Cliffs, NJ.
Black, C. (ed.) Methods of Soil Analysis – Part 1: Physicaland Mineralogical Properties, Including Statistics ofMeasurement and Sampling, 17th edition. American Societyof Agronomy, Inc, Madison, WI.
Calder, N. J., Anderson, B. C. & Martin, D. G. Fieldinvestigation of advanced filtration for phosphorus removalfrom constructed wetland effluents. Environ. Tech. 27 (10),1063–1071.
Clesceri, L., Greenberg, A. & Eaton, A. Standard Methods forExamination of Water and Wastewater. American PublicHealth Association, Washington, DC.
Diamond, D. Determination of Orthophosphates in Watersby Flow Injection Analysis Colorimetry. Quikchem Method10-115-01-1-A. Revised 15 March 2001, Lachat Instruments,Milwaukee, WI.
Drizo, A., Frost, C. A., Grace, J. & Smith, K. A. Phosphateand ammonium removal by constructed wetlands withhorizontal subsurface flow using shale as a substrate. WaterSci. Technol. 35 (5), 95–102.
Drizo, A., Frost, C. A. & Grace, J. Physico-chemical screeningof phosphate-removing substrates for use in constructedwetland systems. Water Res. 33 (17), 3595–3602.
Drizo, A., Frost, C. A., Grace, J. & Smith, K. A. Phosphateand ammonium distribution in a pilot scale constructedwetland with horizontal subsurface flow using shale as asubstrate. Water Res. 34 (9), 2483–2490.
Drizo, A., Forget, C., Chapuis, R. P. & Comeau, Y. Phosphateremoval by EAF steel slag – a parameter for the estimation ofthe longevity of constructed wetland systems. Environ. Sci.Tech. 38, 892–898.
Drizo, A., Forget, C., Chapuis, R. P. & Comeau, Y. Phosphateremoval by electric arc furnace steel slag and serpentinite.Water Res. 40 (8), 1547–1554.
Faulkner, S. P. & Richardson, C. J. Physical and chemicalcharacteristics of freshwater wetland soils. In: ConstructedWetlands for Wastewater Treatment (D. A. Hammer, ed.).Lewis Publishers, Chelsea, MI, USA, pp. 41–66.
Garcia, J., Aguirre, P., Barragan, J., Mujeriego, R., Matamoros, V.& Bayona, J. Effect of key design parameters on theefficiency of horizontal subsurface flow constructedwetlands. Ecol. Eng. 25, 405–418.
Gray, C. A. & Schwab, A. P. Phosphorus fixing ability of highpH, high calcium, coal-combustion, waste materials. WaterAir Soil Pollut. 69 (3–4), 309–320.
Johansson, L. & Gustafsson, J. P. Phosphate removal usingblast furnace slag and opoka – mechanisms. Water Res.34 (1), 259–265.
Kadlec, R. H. & Knight, L. Treatment Wetlands. LewisPublishers, Boca Raton, FL.
Kirk, D. W., Jia, C. Q., Yan, J. & Torrenueva, A. L. Wastewater remediation using coal ash. J. Material CyclesWaste Manage. 5, 5–8.
Laska, M. An Evaluation of FreshWater Mollusc Shell Sand asan Adsorptive Media for Removing Phosphorus in ConstructedWetlandsFilter.M.Sc. Thesis,Department ofCivil Engineering.Queen’s University, Kingston, Ontario, Canada.
Mann, R. A. & Bavor, H. J. Phosphorus removal inconstructed wetlands using gravel and industrial wastesubstrata. Water Sci. Technol. 27 (1), 107–113.
Metcalf & Eddy Inc. Wastewater Engineering: Treatment andReuse, 4th edition. McGraw-Hill, New York.
Ontario Ministry of Environment. Environmental Permits andRegulations. Environmental Protection Act –O. Reg.560/94. Available on-line at: www.e-laws.gov.on.on/DBLaws/Regs/English/940560_e.htm.
Orive, E., Elliot, M. & de Jonge, V. N. Nutrients andEutrophication in Estuaries and Coastal Waters(Developments in Hydrobiology). Kluwer AcademicPublishors Group, MA, USA.
Prochaska, C. A. & Zouboulis, A. I. Removal ofphosphates by pilot vertical flow constructed wetlands usinga mixture of sand and dolomite as substrate. Ecol. Eng. 26,293–303.
311 A. S. Paul & B. Anderson | Phosphorus removal by blast furnace slag and cement clinker Water Quality Research Journal of Canada | 46.4 | 2011
Reddy, K. R. & D’Angelo, E. M. Biogeochemical indicators toevaluate pollutant removal efficiency in constructedwetlands. Water Sci. Technol. 35 (5), 1–10.
Reed, S. C. & Brown, D. S. Subsurface flow wetlands – Aperformance evaluation. Water Environ. Res. 67 (2),244–248.
Richardson, C. J. & Craft, C. B. Effective phosphorusretention in wetlands: Fact or fiction? In: ConstructedWetlands for Water Quality Improvement (G. A. Moshiri,ed.). Lewis Publishers, Boca Raton, USA, pp. 271–282.
Rosolen, S. An Evaluation of Adsorptive Media forImproving Phosphorus Removal in Constructed TreatmentWetlands. M.Sc. Thesis, Department of Civil Engineering,Queen’s University, Kingston, Ontario, Canada.
Seo, D. C., Cho, J. S., Lee, H. J. & Heo, J. S. Phosphorusretention capacity of filter media for estimating the longevityof constructed wetland. Water Res. 39 (11), 2445–2457.
Shilton, A. N., Elmetri, I., Drizo, A., Pratt, S., Haverkamp, R. G. &Bilby, S. C. Phosphorus removal by an ‘active’slag filter: a decade of full scale experience.Water Res. 40 (1),113–118.
Toet, S., Van Logtestijn, R. S. P., Kampf, R., Schreijer, M. &Verhoeven, J. T. A. The effect of hydraulic retentiontime on the removal of pollutants from sewage treatmentplant effluent in a surface-flow wetland system. Wetlands 25,375–391.
Van Weelden, T. Phosphorus Adsorption Capacity of ZebraMussel Shells as a Potential Filter Medium in ConstructedWetland Systems. B.Sc. Project, Department of CivilEngineering, Queen’s University, Kingston, Ontario, Canada.
Vomicil, J. A. Methods of Soil Analysis – Part 1: Physical andMineralogical Properties, Including Statistics of Measurementand Sampling, Porosity, 17th edition. American Society ofAgronomy Inc, Madison, WI, pp. 299–314.
Wood, A. Constructed wetlands for wastewater treatment –engineering and design considerations. In: ConstructedWetlands in Water Pollution Control (Advanced WaterPollution Control no. 11) (P. F. Cooper & B. C. Findlater,eds.). Pergamon Press, Oxford, UK, pp. 481–494.
Xu, D., Xu, J., Wu, J. & Muhammad, A. Studies on thephosphorus sorption capacity of substrates used inconstructed wetland systems. Chemosphere 63, 344–352.
First received 24 February 2011; accepted in revised form 15 November 2011